keptn Community, 15 Oct 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Keptn Community Meeting - October 15th, 2020

Description

October 15, 2020
Diary - what happened last week?
Invalidating evaluation-done events
Release 0.7.2 (by tomorrow)
Image-Pull Policy changed from Always to IfNotPresent
CommunityBridge - Mentoring Program from CNCF: selection has been made! Great to have you onboard the Keptn community.
Hacktoberfest - we are part of it as well as contributing.today
Keptn mailing list / google group: https://groups.google.com/forum/#!forum/keptn
Good first issues: https://goodfirstissue.dev/ or start here
Keptn Community Meetings now included in official CNCF calendar: https://www.cncf.io/calendar/
Noteworthy Issues & PRs (Review)
For each PR, please tell us if you updated the docs, or in case it was not necessary please say so.
Florian
Addressed changes in GitHub regarding default branch: https://github.com/keptn/keptn/issues/2487
Changed ImagePullPolicy to IfNotPresent: https://github.com/keptn/keptn/issues/2518
Christian: FQDN Topic (Follow-ups: https://github.com/keptn/keptn/issues/2551, https://github.com/keptn/keptn.github.io/issues/664 )
Bernd:
Missing triggeredid https://github.com/keptn/keptn/pull/2537
Ankit:

Roadmap → KEP / SPEC / Issues
KEP 06: The next generation of Shipyard: https://github.com/keptn/enhancement-proposals/pull/6
Work in progress:
https://github.com/keptn/keptn/projects/29
Backlog Refinement / Next Sprint:
https://github.com/issues?q=is%3Aopen+is%3Aissue+user%3Akeptn+user%3Akeptn-contrib+label%3A%22next-sprint%22
Planning Board: https://github.com/keptn/keptn/projects/29

Learn more: https://keptn.sh
Get started with tutorials: https://tutorials.keptn.sh
Join us in Slack: https://slack.keptn.sh
Star us on Github: https://github.com/keptn/keptn
Follow us on Twitter: https://twitter.com/keptnProject

A

Okay, hi everyone welcome to this to this captain community meeting. I hope you can see my screen. I just did the screen sharing uh all good um and yeah it's october 15th. um We are um pretty much uh in uh releasing a new version of captain. I guess we will hear a little bit more uh later on um what I just want to mention a couple of things in the beginning, again, our housekeeping rules.

A

We are a cncf project, so we are following the cncf code of contact uh that basic means be nice to each other, um in the chat and also on audio and on video.

A

We see that we're very happy to see our community growing um and especially in the community bridge, that's a program, a mentoring program from the cncf. So we are happy to have two mentees for the next couple of weeks um in this mentoring program and uh they are working on uh basically two uh issues on improving capital. So we are really happy um to see this happening. um Then captain is also part of the hacktoberfest.

A

uh We submitted a couple of uh issues to the hacktoberfest, so if you are interested in contributing to an open source project um feel free to do so. We are really happy to accept contributions uh in this weekly calls uh right now uh like in these calls like this. um We have uh a huge part of the captain development team here, so you can also ask questions or present your contributions and um on github you will find under the search term hacktoberfest you will find also captain.

A

Our issue is also tagged with oktoberfest, so we will. We are happy to to see some contributions there um and then one other part, um the captain community meetings. uh They are now added also to the cncf to the official cncf calendar. So if you take a look at the cncf calendar and maybe you're interested also in other open source projects, there are a lot of great projects out there.

A

You can subscribe to this calendar and you will also see the captain community meetings just make sure when you click on this, uh that it reflects also your time zone. So, in my case it doesn't- and it doesn't do a good job here. um But if you go to more details or copy to my calendar, then you will see it in your calendar and then it's the correct time. So actually that should be the correct time. It's just, I think the utc and not the c plus two.

B

It looks like that first day is a busy day for cncf projects. Yeah.

A

um Thursday 3 p.m. Yeah we are, we are in uh uh yeah in competition with uh the vitas monthly meeting, but for example, next week, 3 pm is all hours so uh yeah, but you can see it's a lot of community meetings. A lot of uh sikh meetings from the cncf uh and also captain now is part of this. um So make sure you have this in your calendar and never miss any captain. Community meetings um yeah the other parts.

A

uh We pretty much talk about them every week, so for this week I will skip them and I already will hand over uh to florian to uh talk a little bit about his um improvements to captain in the last week and uh again, if there is someone uh in the audience who also wants to share the screen, please just let us know in the chat um or in the q a section you can ask the question, so we will try to address everything during this meeting.

C

All right, thank you, jung. You should see my screen already.

C

All right uh today, I want to show you uh two issues I have worked on during this week. The first one basically addresses the recent changes that github introduced about their repositories.

C

So you might have heard that since the beginning of october, if you create a new github repository, the main or the default branch will not call will not be called master anymore, but now it's going to be called main. It's a change that github made to be more inclusive, but unfortunately, this uh led to problems for uh projects that were created in captain with a git hub upstream repository where that already was the case, and therefore we have added a fix for that for the upcoming captain version, which is going to be zero 7.2.

C

So, basically, what we did here was we basically just made sure that the master branch exists and if no master branch is there on the upstream repository, we're gonna, create it, and then everything will be working as.

C

As you're used to as a user, so from a user perspective, nothing will change. So this is the fix I've done for the upcoming version, because this is done this way, because we didn't want to introduce too many changes in the logic of how our configuration service communicates with the upstream repository.

C

So that was the safest approach that we could do within this short time time frame, because this issue arose during our release process, but for the next version for 0.8, I have implemented a different approach where I actually, or with a configuration service, actually uh retrieves. The the information about the upstream repository using the git, show remote command and in the output of that command it will get the information about the default branch.

C

So in that case, on a github repository, this will tell the configuration service that the default branch is called main and then the configuration service uses that name of that branch to cro, to create all the other branches and do the do the communication with the upstream repository all right.

C

Are there any questions about that.

A

I would have a question on this, um so for me, as a user. uh That means, if I do this to let's say tomorrow, with the new release, um there will be a main and a master branch for every capital repository and then starting with captain 0.8, uh the the master will be gone. There will be only main and then this uh the different branches for different stages, exactly yeah. So.

C

For the version that's going to be released in the next days, the master branch will be created if it's not there, but then in the next versions. This will not be the case anymore. So from there on it will not create a master branch. If a user chooses to use a different name for the default branch of the upstream repo.

D

I've learned just one quick note: uh please don't let me forget that we need to remove the warning that we put in place on the on the documentation, because there is now a warning that says that there must be a master branch, but with your with your improvement, we can remove this warning.

C

A

I also have one suggestion: uh if you're doing this on github, then normally on github, you can define one master or main branch like the default branch and that's always set to master or main. So probably, I guess right now. It's set to main, but if we're creating the master branch, the main will be empty.

A

So there could be something that we can give a hint to the user. If you're using this uh and the main branch is it's empty, but it's always defaulting to the main branch that probably sets your default branch to the master so that you can see something.

C

All right, if there are no further questions about that issue, I want to continue with the next one. So this is also a change. That's uh going to be included in the version that's going to be released within the next few days. The 072..

C

This is a very small change, but uh an important one. So we noticed that we use the image pull policy, always in our deployments in the helm, charts uh of the or in a captain, uh installer hem chart, and this imaginable policy might lead to failures. If you are, for example, in an air gap system and therefore we have changed it to, if not present- and this should help to remediate or to mitigate these problems.

C

Any questions about that.

C

All right, if not, I will hand over to place down.

B

Thanks so uh I only have a small topic today to talk about not even a pull request, but something that's ongoing.

B

We noticed a couple of problems this week with name server lookups within the kubernetes cluster and in fact they are not related to captain at all, but we want to discuss them briefly. In case you run into the same error.

B

There will also there's also going to be an advisory on github soon or at least the troubleshooting section on our docs soon so I'll. Take that on my agenda to create the issues in the troubleshooting section. So uh basically, what did we see? We saw error messages. I hope you can read them in the prometheus sli service and also in other services that basically say something like the lookup of eventbroker.captain.service cluster local failed.

B

Sometimes it's also not the event broker. It could be any service within our cluster, but we usually access them using the fully qualified domain names in the communities cluster which are made up, which consists of the service name, the namespace, the service lives in and service.cluster.local.

B

So um that kind of thing failed on us on several occasions now um and we were like really puzzled. What is going on and um thank you to thomas schutz, who isn't here right now, but big. Thank you to him. He helped us in troubleshooting this and we found out that it had something to do with the local name, server configuration or rather more linux's resolve.conf file and the way that kubernetes uses it in certain instances.

B

So let's describe the symptom first for an instance, if I do a wget on I'm logged into my communities cluster right now, so this is a port in my cluster and I just do a wget eventbroker.captain.service cluster local.

B

This is what happens. I have to wait a little bit and essential. Eventually I get an error message saying bad address and I'm like this is not a bad address. This is an address that cuban leaders should understand. But, okay, I'm an engineer. I can go further and I say give me the ip address of that service using nslookup and usually it works.

B

So maybe you're like me, and you know like okay, I don't know what's going on, then you go to an engineer that knows a lot of candidates like thomas, and he will tell you that this is something to do with your resolve, conf, which is inherited on the cluster and yeah. Essentially, what we have in here.

B

We need to show this right now. We have a search domain path in here, and one of these is just not valid or returned something that's not good, and that only happens in combination with options n.5, which is something that cumulative sets by default yeah, but there's not much. We can do about that, except for removing some search domains or removing that setting, but that's something the server administrator needs to do. That's not nothing that captain can do, but uh essentially this is the way you can find out.

B

If you get something like that, uh whether you're affected another way to find out. This is also something I want to show you you can test it by using this command.

B

Essentially, I'm going to copy it here, we're using cube ctl to run an alpine 3.11, because that's what we base our images on and within that alpine container we run wget with this domain so event broker.captain service, cluster, local- and in this case this will work, because I have already configured my system properly. The shell that you saw before I had to manually import the wrong configuration, but this should not work there. We go it works and the way you can see it works is you get a correct, http response.

B

It says http 400 bad request, but this is something that was actually uh responded by the event broker in this case. Another way you can try it out if you're not using captain, but if you just want to check if kubernetes is behaving as as normal. You can use this one where you access the kubernetes internal api using kubernetes.default.service cluster local, and it will also reply with an http error unauthorized because you need an api token for kubernetes, but yeah. This error is okay.

B

This is expected, but if you get like bad address or something like that, then yeah, then you might be affected by this problem. As already said, I noticed a couple of people having similar issues. This could be something related to global name servers as well. I'm not sure about that. But if you are affected, um go to your system administrator, ask them about the name: server configuration and double check your etcresolve.conf yeah and, as mentioned I'll I'll write a troubleshooting section. Entry about that.

A

uh We got one question from from ankit from the community, uh which is the wrong search domain, so uh I think it refers to the.

B

A

And you share it again, which one was the correct.

B

In in this case, it was a dynatrace internal search domain, um that's not behaving as it should. um I I'm not fully um involved in the process of that. Yet I just know that this specific domain is the wrong one, um but you can just remove one domain and then try again.

B

I don't want to share the domain right now, because I think that's that's some internal stuff that's going on, but you can check it if you just open the etcresolve.conf and uh remove one search domain and try another one.

B

um Yeah unkit also is just saying suggesting to set n dots to one yep. That's one of the options we are considering to to use it in a later stage. You can override the dns configuration of kubernetes in your deployment, manifest um it's a good input, but we're not 100 sure. If that's something we want to do, uh because it has some impact in on how you work with cuban leaders.

B

But at the same time we've had people report in several blog posts that apparently setting endorse to one is much more performant but I'll open, an issue on that and I'll. We can take the discussion in this issue, I'm pretty sure, there's some better engineers, more qualified engineers that know about the endodonts and dns settings a little bit more than I do, and then we can discuss it. There.

B

All right, um the next one in the list is band. Do you have any pull requests to present or do you want to present any of your pull requests.

E

um Yes, I can actually show you a rather small thing. I've worked on. Let me show my screen.

E

Okay um yeah: this pure request is already emerged and it was dealing with a missing.

F

E

Id field in the sweger definition of the mongodb data store, so this pull request or the changes are not very spectacular. I've just added the triggered id to this workout definition and we generated the server code of the mongodb data store and the screenshots I've added here demonstrates that in the and bridge.

E

Actually this triggered id is now visible and also has the correct value which is uh more or less a link to so here we here we see a approver triggered event which isn't the id and the triggered id value should be the same as the the one from the trigot approval triggered event, which is the case, so I've just added the screenshots to show you that it's not working and that's actually all. I have to say.

B

All right any questions on that pr by anyone else.

B

Okay, then, thank you for your pr um all right, I'm looking at johannes, because you're.

D

B

On the list, do you have anything else we could go.

D

I don't have an issue to talk about, as I was busy with another project, and I was.

F

D

Working on the documentation, but um we can jump over to there to the booking items um we have.

A

Two activities yeah at one thing uh or actually um two things.

D

A

uh I I don't have anything to share yet, but uh just there's a little bit of a spoiler, so uh there will be. uh We are working closely with uh with two contributors on new additions to captain, so there will be a service uh released pretty soon, probably already tomorrow, to um trigger chaos tests from captain, so we are working with the litmus uh community. It's litmus chaos, that's their uh cncf project, we're working with them to also have this included in captain and uh be able to trigger chaos tests from captain.

A

So probably that's something I can show next week uh we're also working uh on integrating enzo tower into captain. That's also something that should be released pretty soon, uh but probably not this week, and one shout out here also or not much out more, like a uh a request. If someone is interested in helping us out in uh improving the captain website, um please reach out to us. um We do have.

A

uh We don't have a guitar issue yet, uh but we will create a github issue for the website uh soon and we are thinking about um reviewing the website. So if someone from the captain community is knowledgeable in css or any um or google, the the framework that we are using or just wants to contribute with some content for the website, we are happy to hear your feedback and yeah just reach out to us. If you, if you are a front-end wizard, uh we are happy to hear your feedback on this uh yeah.

A

So that's just uh some some working items that are in in the process right now and yeah, I'm handing back to johannes, I guess, to jump to the table.

B

I I was just going to ask: we have two attendees um ankit and da yush. If you have any pull requests, you need reviewed or if you have anything that you want to share.

D

In the meanwhile, I was also taking a look into the in reviews. Section and ankit has one pier in there.

A

Okay, so that's great, that's great ankit! I will promote you, so you can also share uh your screen and uh yeah. I think he's just coming back here.

F

F

One pr for fixing the bug and creating shipyard file- uh it's it's http! So before that this, this is being, I think, uh done some. uh I don't remember the name. It is being fixed. My feature requested in this in this issue. I think 1648 uh it is being done by, uh but he he he introduced a bug. While doing this, he left the he left to pass the content body uh of the shipyard file uh to our uh to our create project command so which, which I have taken care of.

F

uh I have to fix this youtube and I think it is being reviewed. I need to cross check the.

F

And there is one more uh issue: removing the distributor from api service and control pane. It is already merged the issue that we have, while using while running the uh while deploying the captain control while deploying the captain humanities cluster, so because these missing distributor topics which puts the foreign flashback which is being fixed in this vr other than that. There is one more issue that is being raised by, I think uh jordan or this, and I think, I'm not sure improve post solution steps. So it is still in pending.

F

State uh question has already left the comments uh for this pr, whether to put whether to expose api, endpoint and api token in the in the command line uh in the shell during travis installation or any other thing. So it is pending for review.

B

Maybe uh we can talk about this.

F

B

Right now um so just to to get everyone on the same page, um the implementation that he did is fine. um He did it according to the what what is what was stated in the issue.

B

My concern now is really: do we want to print the api url and the api token on console when doing a captain install, because that means for any system that uses automation and travis ci is one of them. You will see this in a public log file and anyone can essentially just abuse it if they want to.

F

I think we can remove api token keeping the endpoint is.

D

A

Would it make sense to have something uh to to basically get rid of using cube ctl? uh So during the campaign installation, you usually don't need cube ctl, you do the captain install uh and then for some parts we need cube ctl for some parts. We don't so it would be in general. It would be cool if we can use captain without using cube ctrl.

F

Yeah I have if.

A

F

Is some captain command.

A

We have already captain configure bridge that you it's it's not right. Now, printing, the the the end point to the bridge, but it's printing, the uh the bridge credentials and if we can do something for captain configure endpoint minus output- and we get the end point.

A

So just one improvement to stay a little bit more in the captain universe uh and not having to use too too much of the cubesat parts.

F

Just one thing so cubekittle endpoint output is, is, after the cubot, come on captain or command right. So when we authenticate our captain cli with the cluster with the proper api endpoints and token, uh then we can run the command captain authentic end point output right so export where it will output the endpoint and api token so uh other other than that. If, if we don't want to use cube pedal binary, then we can use cube client to fetch the end point uh for the from the load balancer or notepod other than that.

F

If we don't want to expose api token- and we can uh like uh put the output as it was previously before, but but yeah well well, like captain also uses cube kettle doing help command because it's it's checks for the cup cattle version and everything. So uh I think we we can't actually get it off with cube cattle binary completely right now, uh because it's slightly coupled even with the health command. But I, when I run captain help it checks for the capital, binary or version for the chicken petal liner right.

A

uh I'm not sure if I, if I understand others, do correctly, uh what was the question if, if you run captain help.

F

It will check for the keep getting well. Then.

A

uh Does it check for if cube ctel is installed.

B

I'm not sure, but I think we do have a check if cube cdl is installed at least that's what we used to have. Maybe florian can help me out here.

A

Does the captain cli has any.

F

A

B

A

B

To have it, but we.

C

B

C

But I'm not 100 sure if that's the case currently we'll need to double check on that.

F

So I'll get it.

B

All right, um so how do we continue anyway? How do we continue with dpr? um Should we just hide the token and print the cube ctl command for now, because anything beyond that, I think, is a follow-up issue anyway,.

A

Yeah, uh printing, the uh the token is convenient, but it's a security issue and uh I I think what christian mentioned is correct. We should not print the token uh and do a follow-up issue on the other parts.

F

Let us make the changes accordingly,.

A

F

Yeah these, these are the following issues that I want to highlight thanks. Thank you.

D

Yes, thanks for contributing cool all right yeah then I would is someone else in the call who wants to present something.

A

I see we have more people joining. So if you want to present something, please indicate in the chat or in the q a but yeah. Maybe we can already uh switch to the next parts and if there's something.

D

Then I will switch over to our working items um that one we currently discussed.

D

That's yeah this one we we talked about. Then. Let's have a look into the issues that are ready for review uh christian. I gave you feedback on on that one.

B

I saw that I still need to work the the feedback into the pull request.

D

B

I'll see if I can get it done tomorrow: okay, thanks.

D

For the feedback.

B

D

Welcome then uh florian you were, you were also working on an issue for invalidating quality gate evaluation results this week. Are you going to present it next week.

C

uh Yes, so it's pretty much done and implemented. uh I just wanted to wait with the pr until after we've released version 072, so it doesn't get uh merged uh into the current release. Branch accidentally makes totally sense. Yeah.

D

Okay, then we leave it in there.

C

Shown this one, but I think there is the pr for the master branch that is not reviewed yet.

D

Use the default branch of origin. Yes, this is the pr for the master branch um yeah, but I think during the next week we will get it reviewed and then then we can close this issue as you presented it um today,.

D

Okay, then uh airmen is not on the call today. He has three pull requests and three issues to to show us next time, and we talked about that one, which is about the improvement of the post installation command, post installation steps for the um installation command all right yeah. I don't want to go into much details about those items that are in progress or assigned and committed. If someone wants or needs feedback, please raise your hand and let us know.

D

The only thing you want to point out that that we are currently finalizing uh the release. 0.72, as you can see here, 123 uh checkboxes are are closed and a couple of of issues are still or a couple of tasks still open, but um we are going to to release.

B

It very soon, actually, almost all of them are done, the ones that are still open uh like they. We ignored them or skipped them for reasons, and we only need to really really finish. The release, write, release, notes and stuff like that, and that's it just to give you an idea where we are.

D

You can scroll down, I will scroll down. Those are the checks boxes that are not uh checked, because it wasn't necessary to release the kubernetes utils and when we go down, we can see that captain officially is the binaries. Are there for the for the entire captain, also the cli and the helm installer um is is provided, and now we we are getting to the final steps.

D

All right, yeah stay tuned and uh try it out on the weekend, then other than that the other documentation is going to be updated as well. And then we have a couple of issues that are in progress.

D

Then we'll also skip the assigned and committed column and we'll directly jump to the backlog. Where we have issues that we want to work on on the next week and the first one is explaining the requirements for the execution plan service and depth in a documentation.

D

And maybe, as a quick reminder, let me just show you one slide to make it more obvious what this change and the documentation is also focusing on.

D

As we talked about the shipyard in version 0.2, which is the next version of the shipyard file, I explained you that, with this new version, we will have here in the shipyard, a sequence in a stage and a sequence then consists of multiple tasks and those tasks are executed. Sequentially.

D

This means that when this sequence gets triggered, then the deployment task is triggered after the deployment, the test then the evaluation. Then there is an approval task and finally, the release task and what the shipyard controller we will do. I will send out an sh.cap event. Deployment triggered and the shipyard controller is waiting for two events.

D

The first one is deployment.finished and there is also- and the shipping control controller also waits for the deployment he waits for it started as well as for the finished event, and this is what we need to consider in the documentation. When someone writes a contribution or a service for the execution plane of captain, then this service must behave like that. This service must listen on a triggered event.

D

He responds with us started and finally, he also responds with a finished event.

D

Yeah, this is what what need to be written down um in our documentation.

A

uh Be honest: how would that the already uh services.

E

um For example,.

A

When we already have the radiation action triggered and the revelation action finished, they already seem to implement. This uh is this for the remediation part? Is this already done, or would we have to change them.

D

Yeah no for the remediation part, we are almost on a good side.

D

There is just one other thing I want to point out later, but over all the other execution services. We need to touch them.

F

D

That is uh yep 0.8 will be a breaking change when it comes to eventing mechanism.

A

And that also affects the sli services, correct.

D

D

All right, um yeah is someone up for writing the documentation for this new feature.

D

Otherwise, I will take it on my.

D

E

D

To yourself, okay,.

D

And as we are currently discussing this topic, let me jump directly to another issue which is called guide for using new captain cloud events in captain contrib and captain sandbox.

D

This issue is a kind of meet the issue because it's just a summary of what we have to do to make all contributions and sandbox projects work with captain 0.8 and he have written down all reports that we need to touch with the main focus on the repos that are in captain contrib. But I also included.

D

Relevant ones that are in sandbox and what you can see here is actually just a list of pull requests in the in the different repositories and what we need to work on is. First of all, we need to upgrade each service to use the cloud events, 0.1 specification and cloud event. 0.1.0.

D

Library- and we also need to make the service- we change the behavior of the service in the sense of listening to triggered event and responding with a started and finished event. For example, we need to do this for the dynadress sli service, then for the premises, sli service and so on and yeah. This is when someone wants to actually learn how an executing service is working.

D

I think this is actually a good starting point, because there you get in touch with the events that are, on the one hand, received and then sent back by the service, and when you understand this concept, which is not that difficult, then you're, really on a good level to start your own. To start writing your own captain service and yeah. We are looking for help, especially in for this issue. As you can see, there are a lot of services.

D

We have to touch in the next week or next week, and you can one one note to you, as you have a very good understanding and overview of what external contributions are there.

D

We also should consider adding those to this list to keep track, which ones we have updated and which ones still need support or need a help to get the event thing done.

A

Right now we have most of the external services in captain sandbox. um When you scroll. Is there also a captain sandbox list, or it's only captain, sir captain contrib.

D

For now, it's mainly focused on captain country, but just added the statistics service from the sandbox, but it would make sense to also keep track on on those that are in sandbox and please, let's use this this meta issue for that good idea.

D

All right, yeah and below this made the issue there are, then all the pull requests that are mentioned in this issue here is the update of the cloud events for the dynadris service, the update of the cloud events sdk for the dynadress sli service and so on, and so on.

D

As I said, when someone wants to work uh on a contribution or on a captain service, please feel free to let us know, and then we will assign it to you, I'd like to take the one for the template repository if it's in there it's in there, it's a.

B

D

Bit further, the statistics unleash.

D

Oh, oh, I forgot to add it I'll. Add it: okay thanks but yeah, but it's created still not in this list.

D

All right, then, let's jump back to the the issue on the top. uh The increased test coverage for the helm service.

D

Yeah andreas was doing a lot of work in regard to refactoring the helm service, also to make it 0.8 compliant, and he had some spots not covered with tests, and this issue will focus will be focused on increasing the test coverage of the helm service is someone up for writing unit tests for the helm service.

E

I can take this one.

D

D

All right, then, let's continue with the shipyard controller.

D

D

The shipyard controller also subscribes to to triggered events that are defined in the shipyard and whenever uh formally, we call them um root events, and whenever such a trigger comes in, then the shipyard controller kicks off the execution of a sequence yeah as a just. Let me just jump over to this slide.

D

I'm talking about this trigger event. Whenever the shipbuild controller sees an event of the type of this type that is mentioned here, then it starts to work and what it does is it sequentially executes those.

D

D

All right uh who wants to work on that one? I.

C

Will take that.

D

One thank you very much.

D

D

All right uh yeah, that actually was it. I think from from my end, because here is the the issue that covers all the sub issues that are then listed here. um I think we we don't need to assign them right now.

D

When you have time available, please let me know which of those issues you want to work on and then I will move it over.

B

Is anyone else wanting to work on any other issue, any good first issues or help bonded issues that are not listed in here.

B

Or as always just write, your name or just say so in the issue and we'll assign it to you as quickly as we can.

D

All right, all right: okay, um do we have additional announcements? No, not in this list.

D

Then I think stay tuned about the 0.72 release. That's coming out and yeah when there are no further questions. I think we are done today with the meeting.

F

Okay uh recently, I was trying to use the default hem chart with captain. So with the kept with captain, there is a few requirements like the image or tag name should be in a value.tml file, but with the open source hand, charts there are most of the services will have a nested image and image image, values, keywords and text. So I was not able to like deploy my open source uh gray final hammer with captain. I need to make changes with within the uh template for grifana to onboard it via captain.

F

Would it be possible for me to onboard such applications without manipulating their template.

B

um It's hard to say, but we have a short guide rather short guide what you need to do if you want to deploy something existing, or rather we have like our own examples that we use you. Cannot you basically cannot onboard just a helm chart um right now. There needs to be some specific requirements met.

B

For instance, it's it's one service per helm chart um so, for instance, the cards microservice is one service and one helm chart, and then the cards db microservices another helm chart that's one of the key requirements and then we also have, I think, the requirement with the values yaml with the replica count. If I'm not mistaken,.

B

um That's something you need to manually edit right now, so it's unfortunately not just deploying an existing help chart. It needs to be modified slightly, and if you just have cuban leaders deployment yamas, then you need to obviously convert them into helm, charts before.

F

Yeah so when we use blue green strategy deployment study in that case, like we need image and text particularly to create a new deployment like primary or secondary deployment. So what if I don't want to create any deployment strategies and just need to onboard it directly on my cluster.

D

Then we can bypass this check right now for blue green deployment. You don't have to provide two deployments. You just need to provide once and then captain takes care of creating the the duplicator and creating uh the second one.

F

Yeah yeah: that's what what! But, if I don't want to create a blueprint deployment that then I don't then captain needs not to create like another. Another deployment for me like it can skip the checks where it checks for image and text and replica accounts. It can directly onboard my application with the hamster that I have fielded with or pass it as input.

B

In theory, yes, but there are a couple of things in the background like istio virtual services um that make it a little bit harder. um The replica account also is not just for blue green deployments also for remediation actions, but yeah. I agree um you should not need those checks. If you just do a direct deployment um as johannes is showing you can set the deployment strategy to blue green or to direct in the in the ship yet yeah, but I think the question of ankit was more related to does.

B

He need to have a very specific helm chart if he only wants to use direct deployments, and I I believe the answer is right now you will have to you need you need those changes. It doesn't work without them. As far as I know, but you're right. If it's only a direct deployment, we could probably lift the requirement for those.

B

But I'm not sure if there is no side effects. um The helm service itself is a little bit.

B

Let's say not that simple to explain.

D

And and to some extent, opinionated in the sense of the helm deploy the helmet chart is allowed to contain just one deployment and one service.

B

But we can use this discussion, maybe to formulate a story um to have a help service. That's not restrictive for such deployments, so, maybe so to say an unopinionated helm service, um a secondary helm service, whatever drop-in replacement, something like that, and also with the upcoming changes in the shipyard.

B

So what we do for version 0.8 there it it will be much easier to just include another helm service in the whole workflow, something that doesn't require such restrictive helm, charts. Let's put it like this.

F

Sure, thanks thanks.

D

But as we're discussing this topic, let me just share you one thought about that: one additional thought, because there is the enhancement proposal, it's number 23.

D

That slightly goes into this direction, but it's it's more about how we will deal with blue green deployments in the future, and there is disco discussions going on whether uh we should rely on argo rollouts for that or we should build it by our own and please check out this this captain huntsman proposal, it's just a one one. um It's a written document where you can uh read the idea and please give us feedback, and we are in more than healthy, more than happy to have to to talk on about that.

F

B

But in the end, it totally would make sense if you could really just say: hey captain on board my helm chart.

B

Yes, this could really be like a game changer for many people, if you, if you could use captain like that and not have to worry about any restrictions, whether it's argo rollout in the future or something else yeah that that's just an additional use case, then.

A

A little bit similar to what we've seen with riser.

B

Riser as well, ryzer is an awesome tool. I think in that sense could become the next go-to standard instead of helm. Maybe.

B

B

Addition right, okay, no more questions, then let's wrap it up. I guess so thanks everybody for your contributions and um for your commitments for the next week, and I guess, if there's nothing else to say we will see each other next week.

D

See you perfect thank.

A